As more Chinese companies track and analyze people's online activities for honed marketing and trend predictions, they're becoming more sophisticated in their approach. Deng Zhangyu reports.

What Chinese region hosts women with the largest breasts? The country's netizens might guess, but its biggest e-shopping website, Taobao.com, knows. It's the Xinjiang Uygur autonomous region, according to an infographic the company produced. Taobao Marketplace created the visual representation from online shoppers' data it collected and analyzed to enable bra companies to tailor advertisements to individual women. It's a reminder that data crunchers are mining our every online activity for marketing gold. And this rising industry is becoming more sophisticated. Box office forecaster Entdigital sifted through social networks to correctly predict China's top-grossing director Feng Xiaogang's 1942 would earn 370-400 million yuan ($60-65 million) - far less than the producers' and public's expectations - a month before it hit screens.

"Big Data analytics can target customers, and predict trends and consumer behavior," data analytics investor and Frost and Sullivan principal consultant Wang Yuquan says.

Advertisers agree it's invaluable to honing their marketing to specific consumers.

"We can analyze every consumer based on his online activity to know what he likes," says Yang Jiongwei, CEO of MediaV advertising company in Shanghai.

MediaV's data department employs about 100 workers - a fifth of its total staff. The company uses 400 terminal servers to track Web users' behavior 24 hours a day.

"I don't know who you are, but I know what you do online," Yang says.

"I know your sex, interests, locations, devices used to access the Internet and many other things," he says.

The automated tracking process takes a thousandth of a second, according to Yang.

A user opens a website containing MediaV's code, which downloads cookies to users' browsers and pairs them with a code number assigned by MediaV's terminals.

It then uses algorithms to determine which clients' ads it should display for that particular Web user. The company creates about 1 billion stamps a day, it says.

"I don't need users' names," Yang says. "What I need to know is what they do. I need to know the person on the other side of the browser."

Targeted marketing has been advancing for a long time but Big Data's predictive capabilities, most of which are drawn from trawling social networks, are rapidly emerging.

Entdigital's founder Liu Han worked at Sina Weibo - a micro blog service akin to China's take on Twitter - for four and a half years before founding the box-office forecaster a year ago.

Liu's company collects user data from Weibo, which had more than 500 million registered users by the end of 2012, Renren (China's answer to Facebook), Douban (a community-oriented social network) and search engines.

"We can identify individuals through social networks," Liu explains.

"We can monitor what they post and discuss to predict what kinds of films will be popular during certain periods."

Liu and his team of 10 engineers collect and analyze data on every film's director, lead actors and actresses, fan loyalty and buzz about the film.

"We examine different dimensions - up to 100 - of every user, including time, location, semantics, circle of friends, etc."

Entdigital can now predict revenue for 70 percent of films in Chinese theaters with 85 percent accuracy, Liu says.

"I hope to give efficient guidance to film producers as early as possible this way," he says.

Producers buy the company's analyses to determine their marketing strategies.

While "Big Data" and "data crunchers" are buzzwords in China's dotcom sector, the country's industry is in its infancy.

"The core value big data brings is analytics through quantitative methods, but these methods were, to some degree, ignored in Chinese culture's long history," says Tu Zipei, director of Asia-Pacific Area at KIT Solutions and author of Big Data Revolution.

Because it's in the early stages, it's harder to trust China's data, says Frank Yu, who has worked in information technology for 20 years. The former Microsoft specialist and current CEO of the crowd-sourcing platform Kwestr points out the lack of data from previous years and murky transparency of public organizations makes this trickier.

There are also privacy concerns, which pose a problem in every country.

Tu explains: "Companies monitor consumers' behavior by various digital means for profit. Governments sit atop data mountains and play the role of Big Brother."

Liu says one could theoretically glean all information about an individual. But it would require a lot of money and manpower because the information is scattered among numerous terminals.

Companies are currently more interested in digging through masses of data from tens of millions of people, rather than any one of them.

Wang, who invests in 16 data analytics startups, expects this might change.

"Computers automatically process epic amounts of data. We can stamp individuals with a code of, say, 0001. But we only know what 0001 does and what we should recommend to him. It's beyond our capacity to know who he is."

What does this mean for privacy?

"The safest way to hide a leaf is to put it in a forest," Wang says.

Contact the writer at dengzhangyu@chinadaily.com.cn.

(China Daily 05/12/2013 page1)